EP0828145A1 - Infrared detector and process for manufacturing it - Google Patents

Abstract

The detector includes a Si substrate with two conducting electrodes (13,13) and a sensitive area (10) whose resistivity varies with temperature. The sensitive area is covered by a layer of n and p doped amorphous silicon. A reflector layer (12) directs the radiation transmitted through the area back towards the sensitive layer. The two electrodes are also made of a temperature sensitive material and have the additional role of absorbing the incident radiation. Their shape is optimised to provide the best optical efficiency.

Description

Technical area

The present invention relates to a detector infrared and the manufacturing process thereof. It relates to the field of detectors infrared radiation, and more specifically to resistive microbolometers operating at temperature ambient.

State of the art

Optical radiation detectors are based on various modes of interaction radiation / matter, for example so-called detectors quantum, photoconductive, photovoltaic, directly convert the energy of radiation into free electric carriers. These carriers are collected by adequate means, such as natural polarization or a polarization applied to detector, and constitute the source of the signal electric, function of the illumination received by the detector. Other detectors called bolometers, particularly suitable for infrared radiation, convert energy into heating. The signal product is the image of the detector temperature. The thermal signal / electrical signal conversion is the result of an electrical property, conductivity, dielectric or other susceptibility of a material or assembly of bolometer materials, variable with the temperature. In a very simplified way, the performance of a quantum detector are proportional to the number of carriers collected for a given light. The performance of a bolometer are proportional to the temperature rise ΔT under the effect of a given illumination, modulated by the rate relative variation of the electrical quantity measured G with temperature, ie ΔT.dG / G.dT. For a resistive bolometer of resistance R we are interested in the quantity dR / R.dT, otherwise noted TCR.

Usually infrared wavelengths interesting are close to 4 micrometers or 10 micrometers for terrestrial applications per following the existence of linked transmission windows to the nature of the atmosphere. Unit detectors have lateral dimensions (in the image plane) minimum of a few wavelengths, or a few tens of micrometers for devices suitable for the band 8 to 12 µm. Such devices are then called "microbolometers".

By nature, unlike most infrared quantum detectors that require a cooling, a detector of this type is capable of operating at room temperature, provided that we know how to discriminate overheating due to incident radiation (signal) from various sources ambient noise (thermal, electrical, etc.). This function is the responsibility of the associated circuits and their implementation. Their ultimate performances depend on a few "technological" parameters, which are explained in the description.

Observation of radiation scenes infrared based on microbolometry is practiced advantageously using matrices (or retinas) of NxM unit bolometric points (pixels), associated individually to circuits that generate the operating stimuli and process the signal resulting electric. These circuits are advantageously installed directly, in "monolithic" assembly or "hybrid", below the matrix plane of microbolometers. They are called thereafter "circuit such a matrix is mounted behind a adequate optics which focus the infrared image of the scene observed. The resulting "thermal image" is converted into electrically manipulated signals by the associated circuits.

The power of infrared radiation actually used per unit area to raise the temperature of a unitary bolometer is equal to Pi.Ar.Fr: Pi being the incident power on the total area of a pixel, Ar the relative absorption in the useful area of the pixel, Fr the factor of filling (useful surface / total surface). The temperature T of the useful part (variable in temperature) observable at equilibrium, i.e. at after a sufficiently long time under illumination constant, is equal to Rth.Pi.Ar.Fr; Rth being the overall thermal resistance between the sensitive area (useful) of the bolometer and its environment. The loss thermals for vacuum-mounted devices (negligible convection) are largely dominated by conduction along the mechanical parts that connect the detector to its support (reading circuit).

Instant variation in level of illumination on the pixel considered results in a temperature variation ΔT reached after typically 3τ, (establishment at 95% of the new equilibrium value) with τ = Rth.Cth: Cth being the thermal capacity of the sensitive part.

In order to illustrate the impact of the various technological quantities characterizing the bolometer on the final performance of an elementary point, we can consider (among other possible arrangements) the case of a resistive bolometer continuously polarized under a bias voltage Vpol ( we are interested in the variation of current under the effect of a variation of illumination) and whose reading uses a "chopped" mode, that is to say the projection alternately of the scene to be observed and of a uniform and constant reference temperature (such as the image of the blade of a "chopper"). The "shooting" is done by projecting a scene element on the bolometer (image point) at t = 0. After a time of rise to thermal equilibrium (typically 3τ), the current flowing through the bolometer is integrated for a time t in the form of a charge Q = t.Vpol / R. Then a uniform reference temperature image is projected on the bolometer (chopper blade) and again after 3τ (new temperature balance) the current is integrated in the same way and compared by adequate electronic means to the previous signal in order to quantify the temperature rise of the element, therefore the relative illuminance received. We show that, in this implementation case, the equivalent noise power (CIP), equal to the smallest infrared power detectable by the device, is expressed by: CIP = (1 / Ar.Fr.Rth). (1 / | TCR |). (q / Q) 1/2 . (2ut / Vpol) 1/2 where ut = kT / q, i.e. 25 mV at room temperature. This relation is valid under the assumption that the noise in the bolometer is "white", that is to say independent of the frequency, because related only to the resistance R0 of the bolometer.

So we see that the equivalent power the lower the noise (the detector the more efficient) that the relative absorption is as well as the filling rate and the thermal resistance of the supports, as well as the TCR temperature coefficient. In reading, the charge integrable deserves to be as high as possible, like the bias voltage. This last parameter is generally limited by the increase in various noise factors, especially in the bolometer itself, when the field in the structure increases. The device withstand voltage without excessive noise is therefore an essential quality.

The thermal capacity also defines the bolometer response speed, as such it is also an important technological parameter because it defines, through the product Rth.Cth, the frequency maximum pixel reading (fm = 1 / 6τ in the example considered if t << τ).

Two examples of art making are described in US-A-5,021,663 and US-A-5,367,167.

Figure 1 shows the concept developed in the first reference, figure 2 the one developed in the second. Typically a unitary microbolometer can be subdivided into three main parts: the heart of the device, whose temperature varies with illumination (zone I in Figures 1 and 2). This heart is suspended using supports (zone II). Finally, zones III maintain the whole.

The layout and detail conformation of the different parts I, II and III represented under linear form to simplify the description can particular be adapted to a matrixing according to two dimensions, in order to achieve, with good compactness, photosensitive retinas for imaging. The "suspended" character arises from the need for a thermal insulation, represented by the quantity Rth, very high between the microbolometer and its environment.

• supporter mécaniquement la zone I ;
• connecter électriquement les bornes du microbolomètre aux zones III ;
• isoler thermiquement la partie sensible 10 des zones III.

A unitary microbolometer is typically in the form of a temperature-sensitive element 10 (central zone I), supported by at least one support element (zone II), more generally two, which plays three roles:

• mechanically support zone I;
• electrically connect the microbolometer terminals to zones III;
• thermally isolate the sensitive part 10 from zones III.

• connecter électriquement et mécaniquement l'ensemble I et II sur le support sous-jacent 12 ;
• positionner en altitude le plan du bolomètre, lorsqu'une couche réfléchissante est disposée en surface du substrat pour améliorer l'absorption infrarouge (paramètre Ar).

The function of zones III is to:

• electrically and mechanically connect the assembly I and II to the underlying support 12;
• position the bolometer plane at altitude, when a reflective layer is placed on the surface of the substrate to improve infrared absorption (parameter Ar).

In Figures 1 and 2 the material bolometric is referenced 14. In FIG. 2, we have plus, an electrical insulating layer 15 and a layer conductor "absorber" 16.

It is indeed necessary to provide for a dedicated or integrated structure to ensure optical absorption Ar in the bolometer as close as possible of 1. This objective is achieved according to laws of electromagnetism when the resistance of equivalent layer (all layers placed in parallel) of the microbolometer is equal to 377 Ω / square (absorption non-selective in wavelength λ), and if we take the precaution to place a mirror (reflector) at a distance λ / 4 below the microbolometer plane for return the ray, transmitted through the whole floating, towards the microbolometer, in phase with the incident radius, taking into account the phase rotation π at metallic reflection. Typically this reflector, not shown in Figures 1 and 2, is arranged in surface of the substrate / read circuit 12. Care is taken to build adequate technological means 11 connections of appropriate height to maintain rigidly the bolometer plane in the vicinity of the specified distance. This provision leads to a wavelength selective device (maximum absorption at λ).

As already mentioned, the reading circuit 12, usually made on a silicon substrate according to the usual techniques of microelectronics, is advantageously arranged below the bolometer, especially in the case of structures matrix of NxM points. Its role is on the one hand to supply the polarization stimuli to the microbolometer (application of a voltage or current between two electrodes 13, 13 '), on the other hand to measure the variations in characteristics induced by the heating of the bolometer under the effect of infrared illumination. The entire system is mounted under vacuum, or under low pressure, to avoid convective heat losses, in a housing waterproof with an optical window on the front and electrical connections to electronics peripheral.

In all cases of construction, the area I must also have a thermal capacity Cth as low as possible, which is achieved by limiting the number of layers stacked in this area at the essential, their thickness at least functional, and if possible choosing materials with low specific heat capacity Cp. Zone II must have a maximum thermal resistance Rth, a negligible electrical resistance compared to that of zone I, and a low thermal capacity (in the since it constitutes a "heat sink" for the zone I that we want to see varying in temperature). The zones III must provide electrical continuity between zones II and substrate 12, i.e. a negligible electrical resistance and resistance sufficient mechanical to hold in position the floating assembly I + II.

Zone I essentially comprises element 10 sensitive to temperature. She understands typically two conductive electrodes 13 and 13 'in contact with a "bolometric" material, the resistance or other electrical characteristic varies with temperature.

These electrodes can be parallel to the substrate plane, above and below the material bolometric which is sandwiched. This configuration is illustrated in Figure 1. According to this first concept, the current resulting from stimuli imposed by the reading circuit crosses the structure perpendicular to the plane of the device. In that case these electrodes are "full", that is to say occupy the all or most of the surface of the zone I. The bolometric material can be silicon lightly doped n or p amorphous, or any material of which conductance, or dielectric susceptibility, or polarization (since the configuration is clearly that of a capacitor) varies with the temperature, for example near the temperature ambient.

In this first configuration the overall layer resistance to optimize coupling optical (Ar absorption) is obtained by adjusting that of the two electrodes 13 and 13 'in the vicinity of 750 Ω / square, since we have two conductors in parallel to the optical sense.

In another configuration, illustrated by Figure 2, the electrodes 13 and 13 'are coplanar and constitute simple lateral contacts, located on the along two opposite edges of zone I and do not occupy typically only a small part of this area. The current in the structure then flows in parallel in terms of the device. This second configuration is better suited to the use of amorphous silicon lightly doped n or p as bolometric material because the electrodes are, by construction, much more distant. However, optical coupling (Ar) cannot be obtained with amorphous silicon alone, the resistivity is much too high. This coupling optics is then obtained by adding a layer conductor 16 specific 377 Ω / square, insulated electrically from the body of zone I by a layer dielectric 15.

• Domaine d'application limité aux matériaux de forte résistivité, typiquement > 5.10⁵ Ωcm pour procurer une résistance Ro de l'ordre du MΩ (0,1 µm de matériau thermosensible, surface 50 µm × 50 µm). Il en résulte, au moins dans le cas du silicium amorphe, des difficultés certaines de contrôle de procédé d'élaboration.
• Les structures de ce type présentent en général, au-delà de quelques centaines de millivolts de polarisation, des caractéristiques tension-courant non linéaires, voire fortement non linéaires, car l'épaisseur de la couche thermosensible doit être faible (submicronique) pour limiter la capacité thermique de l'ensemble. Ces faibles épaisseurs s'accompagnent ordinairement de phénomènes de conduction dominés notamment par les interfaces, typiquement bruyante et non linéaire. Cette caractéristique liée au très faible espacement des deux électrodes est préjudiciable aux performances d'imagerie de la rétine. Dans certaines conditions, des matériaux comme le silicium amorphe dopé ou des oxydes de vanadium peuvent présenter une linéarité acceptable, mais seulement aux faibles tensions de polarisation Vpol, ce qui tend à limiter les performances en termes de rapport signal / bruit, par affaiblissement du signal.
• Le bruit électrique de telles structures tend ordinairement à augmenter rapidement quand la tension de polarisation augmente ; par apparition d'autres mécanismes de bruit que le "bruit blanc" lié à la valeur de la résistance. Il en résulte que, même si on admet de travailler en zone non linéaire (sous forte tension pour augmenter le signal), le rapport signal/bruit se détériore par augmentation du bruit.
• La réalisation est complexe (cinq niveaux de masquage au moins) car les électrodes 13 et 13' sont indépendantes, ce qui nécessite, quelles que soient les astuces d'assemblage, un niveau lithographique et des opérations associées pour réaliser les reprises de contact dans les zones III, respectivement sur 13 et 13'. D'autre part, au moins trois couches sont nécessaires pour réaliser la structure (électrode 13/matériau bolométrique/électrode 13'), d'où un nombre d'opérations de construction élevé, c'est-à-dire finalement un coût élevé (nombre d'opérations / rendement).
• Certains procédés présentent des caractéristiques critiques vitales, qui jouent sur le potentiel de rendement, par rapport à la fonctionnalité du dispositif : gravure des connexions situées dans les zones III, gravure de l'électrode 13', gravure périphérique de réticulation, intégrité des couches très fines prises en sandwich entre deux conducteurs non équipotentiels. Toutes ces caractéristiques sont potentiellement la source de fuites électriques entre électrodes, elles-mêmes sources de perturbations du signal (dispersions de niveau en structure matricielle) ou/et de bruit excédentaire.
• La fonction absorbeur infrarouge nécessite de réaliser deux couches de métal avec des résistances de couche de 750 Ω/carré, difficile à réaliser et contrôler avec les métaux usuels, qui doivent par ailleurs présenter de bonnes caractéristiques de contact sur le matériau bolométrique. Le nitrure de titane (TiN) par exemple présente des résistivités élevées, 100 à 300 µΩcm usuellement suivant le procédé utilisé. Les épaisseurs requises vont de 13 à 40 Angström suivant la résistivité, et beaucoup moins avec les métaux usuels (Titane, Chrome...) .
• Le dessin des zones d'interconnexion III occupe une surface relativement importante (deux vias+gardes et espaces divers) au détriment du facteur de remplissage Fr.

The concept illustrated in FIG. 1 (US-A-5,021,663) has the following drawbacks:

• Field of application limited to materials with high resistivity, typically> 5.10 ⁵ Ωcm to provide a Ro resistance of the order of MΩ (0.1 µm of heat-sensitive material, surface 50 µm × 50 µm). As a result, at least in the case of amorphous silicon, certain difficulties in controlling the production process.
• Structures of this type generally have, beyond a few hundred millivolts of polarization, non-linear, even highly non-linear voltage-current characteristics, because the thickness of the heat-sensitive layer must be low (submicron) to limit the thermal capacity of the assembly. These small thicknesses are usually accompanied by conduction phenomena dominated in particular by the interfaces, typically noisy and non-linear. This characteristic linked to the very small spacing of the two electrodes is detrimental to the imaging performance of the retina. Under certain conditions, materials such as doped amorphous silicon or vanadium oxides may exhibit acceptable linearity, but only at low polarization voltages Vpol, which tends to limit the performance in terms of signal / noise ratio, by signal weakening .
• The electrical noise of such structures usually tends to increase rapidly as the bias voltage increases; by the appearance of noise mechanisms other than "white noise" linked to the value of the resistance. As a result, even if it is accepted to work in a non-linear zone (under high voltage to increase the signal), the signal / noise ratio deteriorates by increasing noise.
• The production is complex (at least five masking levels) because the electrodes 13 and 13 ′ are independent, which requires, whatever the assembly tricks, a lithographic level and associated operations to carry out contact recovery in the zones III, respectively on 13 and 13 '. On the other hand, at least three layers are necessary to make the structure (electrode 13 / bolometric material / electrode 13 '), which results in a high number of construction operations, that is to say ultimately a high cost. (number of operations / yield).
• Certain processes have vital critical characteristics, which affect the yield potential, in relation to the functionality of the device: etching of the connections located in zones III, etching of the electrode 13 ', etching of the crosslinking device, integrity of the layers very fine sandwiched between two non-equipotential conductors. All these characteristics are potentially the source of electrical leaks between electrodes, themselves sources of signal disturbances (level dispersions in matrix structure) or / and excess noise.
• The infrared absorber function requires making two layers of metal with layer resistances of 750 Ω / square, difficult to achieve and control with common metals, which must moreover have good contact characteristics on the bolometric material. Titanium nitride (TiN) for example has high resistivities, 100 to 300 μΩcm usually depending on the process used. The thicknesses required range from 13 to 40 Angstroms, depending on the resistivity, and much less with common metals (Titanium, Chrome, etc.).
• The design of interconnection zones III occupies a relatively large area (two vias + guards and various spaces) to the detriment of the filling factor Fr.

• L'espacement élevé des électrodes du bolomètre, égal pratiquement au pas de répétition en structure matricielle, à épaisseur constante de matériau thermosensible (cette épaisseur est dictée par la contrainte sur Cth, c'est-à-dire minimum) implique une résistivité de ce dernier relativement faible pour assurer un niveau de résistance R0 adapté au couplage avec le circuit de lecture. Dans le cas du silicium amorphe, cela signifie un coefficient dR/RdT relativement faible. Cette approche est donc limitée en performances.
• La fonction absorbeur infrarouge nécessite, du fait de l'espacement très élevé entre les deux électrodes, une couche métallique sur toute la surface sensible, doublée d'une couche d'isolement diélectrique pour éviter le court-circuit avec les électrodes actives. L'épaisseur totale résultante de matériaux empilés est élevée (1500 à 2500 Angström de matières diverses), d'où une capacité thermique importante de la zone sensible.
• La statistique d'isolement (c'est-à-dire l'intégrité physique) de la couche de passivation (nécessairement très fine) entre absorbeur infrarouge et matériau sensible à la température, définit le rendement fonctionnel des dispositifs. Le dépôt de cette couche est donc une étape critique.
• La réalisation est complexe (six niveaux de masquage).
• Les interconnexions vers le circuit de lecture nécessitent une surface non négligeable (deux vias + un espace entre connexions). Le taux de remplissage Fr en est diminué d'autant.

The concept shown in FIG. 2 (US-A-5,367,167) has the following drawbacks:

• The high spacing of the bolometer electrodes, practically equal to the repetition step in a matrix structure, with a constant thickness of thermosensitive material (this thickness is dictated by the stress on Cth, i.e. minimum) implies a resistivity of this the latter relatively low to ensure a resistance level R0 suitable for coupling with the read circuit. In the case of amorphous silicon, this means a relatively low coefficient dR / RdT. This approach is therefore limited in performance.
• The infrared absorber function requires, due to the very high spacing between the two electrodes, a metal layer over the entire sensitive surface, doubled with a dielectric insulation layer to avoid short-circuiting with the active electrodes. The resulting total thickness of stacked materials is high (1500 to 2500 Angstroms of various materials), resulting in a significant thermal capacity of the sensitive area.
• The isolation statistics (that is to say the physical integrity) of the passivation layer (necessarily very thin) between infrared absorber and temperature-sensitive material, defines the functional performance of the devices. The deposition of this layer is therefore a critical step.
• The realization is complex (six levels of masking).
• The interconnections to the reading circuit require a considerable surface (two vias + a space between connections). The filling rate Fr is reduced accordingly.

Compared to resistive bolometers current, who do not have a relationship linear current / voltage, high voltage, and naturally noisy (electrode devices parallel) or have, in addition to their electrodes, a metallic layer in another plane to ensure the radiation absorption function (devices with coplanar electrodes), the subject of the invention is a bolometer-type infrared detector resistive to coplanar conductive elements which ensure the function of electrodes and the absorption of radiation while retaining the linearity of the current / voltage relationship and low noise level even at high voltage.

In addition, this invention allows the implementation more efficient bolometric materials (more resistive) following a technological assembly simpler.

Statement of the invention

• une partie sensible comportant :
  • un élément sensible comportant une couche de matériau dont la résistivité varie avec la température,
  • des éléments conducteurs permettant d'assurer les fonctions d'électrodes dudit détecteur et d'absorbeur infrarouge ;
• au moins un élément de support de la partie sensible apte à positionner ladite partie sensible, à la relier électriquement à un circuit de lecture et à l'isoler thermiquement du reste de la structure ;

caractérisé en ce que tous les éléments conducteurs sont disposés sur la même face de la couche de matériau sensible à la température. The present invention relates to an infrared detector comprising:

• a sensitive part comprising:
  • a sensitive element comprising a layer of material whose resistivity varies with temperature,
  • conductive elements making it possible to ensure the functions of electrodes of said detector and of infrared absorber;
• at least one support element for the sensitive part able to position said sensitive part, to connect it electrically to a reading circuit and to thermally isolate it from the rest of the structure;

characterized in that all the conductive elements are arranged on the same face of the layer of temperature-sensitive material.

According to a first embodiment, the conductive elements have two electrodes interdigitated simultaneously performing the functions electrodes and absorbers, these electrodes being connected to the reading circuit.

According to a second embodiment, the conductive elements have two electrodes of a first type connected to the reading circuit and at least a second type electrode brought to a potential floating, the electrode of the second type ensuring only the absorber function and the electrodes of the first type simultaneously performing the functions electrodes and absorber.

Advantageously at least two elements conductors provide the electrical connection to the reading circuit.

According to a first variant, said face of the layer of sensitive material is the underside, the conductive elements being arranged under said layer.

According to a second variant, said face of the layer of sensitive material is the upper face, the conductive elements being arranged on said layer.

According to a third variant, the elements conductors are located between two layers of material sensitive, i.e. in contact with the face top of a first layer and the underside a second layer of sensitive material.

Advantageously, the material sensitive to temperature is doped amorphous silicon. He can be also a vanadium oxide.

Advantageously, the free space defined between the conductive elements consists only of temperature sensitive material.

Advantageously, a reflector is provided in surface of the substrate supporting the reading circuit, and located at λ / 4 below the plane of the sensitive element, λ being the wavelength of the radiation R for which the bolometer will preferably be sensitive.

Advantageously the pattern of the elements conductors is repeated in steps typically between λ and λ / 2 to obtain a average absorption greater than 90%; for example for detection in the 8-12 µm band, the electrode pitch is less than or equal to 8 micrometers, to avoid the cutoff effect of the “short” wavelengths (8 at 10 µm) and greater than or equal to 5 µm to avoid polarization effects.

Advantageously, each support element has at least one connection pillar and at least one thermal insulation arm.

The invention also relates to a detector consisting of a set of unitary detectors such as defined above.

a) formation d'une couche sacrificielle ;

b) dépôt d'au moins une couche conductrice pour la réalisation des éléments conducteurs ;

c) réalisation d'au moins une ouverture de contact vers le circuit de lecture par gravure localisée de la couche conductrice et de la couche sacrificielle ;

d) dépôt d'au moins un matériau conducteur et gravure de ce dernier de façon à assurer la liaison électrique entre les éléments conducteurs et le circuit de lecture, le motif ainsi réalisé formant le pilier du support ;

e) gravure de la couche conductrice déposée à l'étape b) de façon à définir les motifs des éléments conducteurs ;

f) dépôt sur l'ensemble, de la couche de matériau sensible à la température et gravure des différentes couches disposées au-dessus de la couche sacrificielle de façon à définir l'élément sensible et le bras du support ;

g) gravure de la couche sacrificielle de façon à libérer la partie sensible, et le pilier.

The invention also relates to a method for manufacturing an infrared detector, characterized in that, starting from a reading circuit already completed from a substrate, it comprises the following steps:

a) formation of a sacrificial layer;

b) depositing at least one conductive layer for producing the conductive elements;

c) making at least one contact opening towards the reading circuit by localized etching of the conductive layer and the sacrificial layer;

d) depositing at least one conductive material and etching the latter so as to ensure the electrical connection between the conductive elements and the reading circuit, the pattern thus produced forming the pillar of the support;

e) etching the conductive layer deposited in step b) so as to define the patterns of the conductive elements;

f) depositing on the assembly, the layer of temperature-sensitive material and etching of the various layers arranged above the sacrificial layer so as to define the sensitive element and the arm of the support;

g) etching of the sacrificial layer so as to release the sensitive part and the pillar.

The order of the previous steps may well heard to be changed, especially according to another embodiment, step f) of material deposition sensitive can be performed before step b), the sensitive material then being etched in step c).

According to yet another embodiment, the sensitive material is deposited according to a first layer before step b) and in a second layer to step f), the first layer of sensitive material then being etched in step c).

Furthermore, depending on the material of the conductive layer of step b), step d) can be performed before or after this step b).

Pillar means any element capable of support, from the support arm, the part sensitive, whatever its geometry.

According to an advantageous mode, one realizes prior to step a), filing and etching a metal layer to form a reflector.

According to an advantageous embodiment the sacrificial layer is a polyimide layer to which was annealed.

According to another advantageous mode of realization, the conductive layer of step b) is a bilayer of titanium and aluminum nitride, the conductive material from step d) being deposited then after completion of steps b) and c), this material conductor and the aluminum then being etched from the same way in step d).

According to another advantageous mode, the material conductor of step d) is a multilayer whose at at least one of the layers is formed by LPCVD.

According to one embodiment, the elements conductors with first and second faces, the sensitive material being in contact with one of the faces of said elements, the other face is placed contact with a layer of passive material.

According to yet another embodiment, a layer of passive material is also deposited on the face (s) of the sensitive material, not in contact with the conductive elements.

Finally, according to an advantageous mode, the passive material is chosen from silicon oxide, silicon nitride and amorphous silicon.

• Cette réalisation présente une simplicité maximale, donc un potentiel de rendement technologique maximum pour deux raisons essentielles :
  • le nombre de couches et donc de niveaux lithographiques pour en définir les contours est minimum,
  • la présence d'une couche conductrice unique (dispositif à électrodes coplanaires sans contre-couche métallique) efface la criticité technologique de tous les matériaux et procédés évoqués précédemment.
• La capacité thermique totale Cth de la partie sensible est minimum grâce à la mise en oeuvre (dans la version la plus dépouillée) nécessaire de deux couches superposées seulement dans la zone sensible : une couche sensible à la température, essentiellement continue, et une couche conductrice n'occupant typiquement qu'une partie de la surface de cette zone.
• Surtout, il est possible simplement à l'aide du dessin des motifs conducteurs de la zone sensible, d'améliorer les performances du détecteur par rapport à l'art antérieur, par l'emploi de matériaux plus résistifs (à TCR plus élevé dans le cas du silicium amorphe), ou d'ajuster la résistance Ro du détecteur pour un matériau de résistivité donnée, ce qui revient également à améliorer le dispositif.

The main advantages of the detector of the invention are as follows:

• This realization presents a maximum simplicity, therefore a potential of maximum technological yield for two essential reasons:
  • the number of layers and therefore of lithographic levels to define the contours is minimum,
  • the presence of a single conductive layer (device with coplanar electrodes without metallic backing) erases the technological criticality of all the materials and processes mentioned above.
• The total thermal capacity Cth of the sensitive part is minimum thanks to the implementation (in the most stripped version) necessary of two superimposed layers only in the sensitive zone: a temperature sensitive layer, essentially continuous, and a conductive layer typically occupying only part of the surface of this area.
• Above all, it is possible simply by using the design of the conductive patterns of the sensitive area, to improve the performance of the detector compared to the prior art, by the use of more resistive materials (at higher TCR in the amorphous silicon), or to adjust the resistance Ro of the detector for a material of given resistivity, which also amounts to improving the device.

• Le coefficient de variation en température TCR peut alors être optimisé, tout en restant assorti d'une caractéristique courant-tension linéaire parce qu'intrinsèque aux dispositifs à électrodes coplanaires. Cette optimisation peut être obtenue par la réduction de l'espace entre les deux électrodes, moyennant certaines précautions relativement à l'absorption optique, comme exposé plus loin, afin d'autoriser la mise en oeuvre de matériaux plus résistifs (à plus fort TCR dans le cas du silicium amorphe), à R0 constant (imposé par Q à Vpol constant). Une autre façon d'exploiter cet arrangement d'électrodes est d'augmenter la tension de polarisation Vpol, à dessin d'électrodes, épaisseur de matériau sensible et Q constants, en augmentant d'autant la résistance R0, c'est-à-dire en utilisant un matériau plus résistif (par exemple par une action adéquate sur le dopage du silicium amorphe), ce qui tend à améliorer TCR dans le cas du silicium amorphe.

A first way of producing these conductive patterns consists in defining two electrodes in the form of interdigitated combs, covering most of the surface of the sensitive area.

• The coefficient of variation in temperature TCR can then be optimized, while remaining associated with a linear current-voltage characteristic because it is intrinsic to the devices with coplanar electrodes. This optimization can be obtained by reducing the space between the two electrodes, with certain precautions relating to optical absorption, as explained below, in order to allow the use of more resistive materials (with higher TCR in the case of amorphous silicon), at constant R0 (imposed by Q at constant Vpol). Another way of exploiting this arrangement of electrodes is to increase the bias voltage Vpol, with electrode design, constant material thickness and Q constant, thereby increasing the resistance R0, that is to say say by using a more resistive material (for example by an adequate action on doping of amorphous silicon), which tends to improve TCR in the case of amorphous silicon.

A second way to achieve metal electrodes involves connecting only two conductive elements occupying only part of the surface of the sensitive area. These two elements constitute the electrodes proper of the device. Other conductive elements are left free to float electrically on either side of these electrodes.

• Accessoirement le facteur de remplissage est optimisé par la mise en oeuvre d'une méthode de connexion électrique et mécanique vers le circuit de lecture de faible encombrement, qui comporte un ou (plus généralement) deux piliers par point élémentaire (ou détecteur unitaire). Certains modes de réalisation permettent la réalisation de connexions à caractéristiques mécaniques améliorées par rapport à l'état de l'art, par disposition symétrique des couches. La conséquence est essentiellement une meilleure statistique de positionnement du plan sensible par rapport au réflecteur, donc une meilleure uniformité de réponse pour un ensemble de détecteurs.

The resistance Ro seen between the two electrodes can vary very significantly (typically by a factor> 60 to 600 for a lithographic definition of 3 µm and 1 µm respectively and a detector size of 50 × 50 µm) between the various ways of making these patterns, with constant thickness and resistivity of the sensitive material. This intrinsic feature of the invention offers a considerable degree of freedom to the designer to optimize the coupling of the detector to the reading circuit, without technological modifications, or adjustment of the resistivity of the sensitive material, possibly inaccessible.

• Incidentally, the filling factor is optimized by the implementation of a method of electrical and mechanical connection to the space-saving reading circuit, which comprises one or (more generally) two pillars per elementary point (or unitary detector). Certain embodiments allow the realization of connections with improved mechanical characteristics compared to the state of the art, by symmetrical arrangement of the layers. The consequence is essentially a better statistic of positioning of the sensitive plane relative to the reflector, therefore a better uniformity of response for a set of detectors.

Brief description of the drawings

• Figures 1A and 1B illustrate a first example of an art detector anteriorly, respectively in a view parallel to the plane of the substrate and in a top view;
• Figures 2A and 2B illustrate a second example of an art detector anteriorly, respectively in a view parallel to the plane of the substrate and in a top view;
• Figures 3A and 3B illustrate the drawing general of the detector according to a first mode of realization of the invention, respectively in a view parallel to the plane of the substrate and in a view of the above ;
• Figures 4A and 4B illustrate two other examples of electrodes of the detector of the invention illustrated in Figure 3;
• Figures 5A and 5B illustrate a second embodiment of the detector according to the invention;
• Figure 6 shows a simulation obtained from Maxwell's equations of the relative absorption Ar as a function of the parameter p / λ (no pattern / wavelength);
• Figure 7 illustrates the types of radiation and the electric field used in the simulation of Figure 6;
• FIGS. 8A, 9A, 10, 11A, 12A, 13, 14A, 15 and 16 illustrate different sections of the structure formed by the detector of the invention, at during the manufacture thereof;
• Figures 8B, 9B, 11B, 12B and 14B illustrate views of different levels lithographic of the detector of the invention, during of making it.

Detailed description of embodiments

The invention consists of a device made with a single conductive layer in which are defined using one level lithographic, patterns corresponding to the elements conductors. According to the invention these conductive elements can be done in two different ways. the Figures 3 and 4 show a first mode of realization of the conductive elements. According to this mode, these elements are produced in the form of two combs interdigitities performing simultaneously the functions microbolometer electrodes, infrared absorber and electrical connection to the reading circuit. This structure applies particularly, but not exclusively, to the use of amorphous silicon lightly doped as a material sensitive to temperature.

More precisely, the the invention given in FIG. 3 comprises an area I comprising two conductive electrodes 13 and 13 'in form of interdigitated combs on the surface, these combs 13 and 13 'can have the shapes illustrated on the Figures 3, 4 and 5.

Zone I of the sensitive part 10 is the temperature sensitive area. It also includes a layer of so-called bolometric material, the resistivity varies with temperature. This layer can be made of n-doped amorphous silicon (a-Si: H) or p.

According to the invention, the two conductive electrodes 13 and 13 ′ in the form of interdigital combs, combine two distinct functions, contrary to the provisions of the prior art: that of electrodes and that of absorption. Furthermore, in this embodiment, it is possible to optimize the free space defined between these two electrodes 13 and 13 ′, which can advantageously consist only of material sensitive to temperature (according to the most stripped embodiment). It advantageously has a minimum length and a maximum width, so as to allow the use of the highest possible resistivity material (at constant R0) to maximize dR / RdT (the length and the width are defined relatively to the direction of the passage current from one electrode to the other). In an elementary point of 50 × 50 µm ² for example, and for a lithographic resolution of two micrometers, it is possible, by simplifying, to define a space of two micrometers in length and (12 x 50) micrometers in width, i.e. overall layer resistance of 1/300 square; the resistance seen between two linear electrodes arranged on two opposite edges of the elementary point equivalent on its side to a square. The gain for a lithography of a resolution micrometer is therefore 300 on the resistivity of the material with equivalent resistance R0 (value under darkness) of the elementary point.

In addition the absorption of radiation infrared received by zone I of sensitive part 10 is a function of the equivalent layer resistance in this zone I and whether or not a reflector on the surface of the substrate 12. It is wise, in accordance with the state of the art, to provide for such reflector to return to the sensitive area the energy which crosses zone I of the sensitive part 10. Under these conditions an absorption is obtained close to 100% in the vicinity of 10 micrometers wavelength, for a space close to 2.5 micrometers (quarter wave) between the reflector and the plane of the sensitive element, and for resistance with a layer close to 380 Ω / square of this same area. Detailed analysis of these optical effects can be found for example in an article by K.C. Liddiard, Infrared Physics, vol. 34, N ° 4, pages 379-387, 1993. Account given the high resistivity of common materials in bolometry, the equivalent layer resistance of the zone I of the sensitive part 10 is equal to that of the (or) conductive layer (s) in parallel which make up. The thickness of this (these) layer (s) must be minimal (optimization of Cth of zone I) and Rth zones II, while remaining technologically practical, that is to say controllable and reproducible. The comb-shaped electrodes according to the invention play this role of infrared absorber in a single layer, advantageously associated with a single level definition lithographic.

The design of the interdigitated electrodes to get the best optical performance possible.

Figure 6 illustrates a simulation made from Maxwell's equations, from the relative absorption Ar as a function of the parameter p / λ (no pattern / wavelength) of an infinite pattern according to the two dimensions normal to the incident ray, consisting of alternating conductive elements width d, spaced from e (p = d + e). This calculation is done for normal incident radiation on the surface of the bolometer, whose electric field is parallel (E //) or perpendicular (E┴) to the large dimension of conductive elements, as illustrated in Figure 7.

Thus in this Figure 7, the simulation assumes one hand the presence of a reflector located at λ / 4 under the plane of the bolometer and an optimum value of the equivalent sheet resistance of the metal pattern Rsquare _eq. # 380 Ω / square .

We will, below, consider successively three possible cases:

First case :

The electric field of the incident radiation is parallel to the large dimension of the patterns (E //).

When p << λ, the pattern behaves as a continuous layer of equivalent square resistance equal to Rcarré éq _. = Square _metal . p / d, and the absorption is close to 100%.

When p = λ the absorption is still worth little almost 80%, then rapidly decreases beyond by diffraction.

Second case :

The electric field of the incident radiation is perpendicular to the large dimension of the patterns (E⊥).

Relative absorption is low for p << λ (extinction by polarization effect), then increases quickly with p, when p> λ / 2, we obtain Ar> 80%; the maximum absorption is reached at p = λ (effect of resonance due to the quarter-wave cavity).

General case :

The orientation of the electric field of the incident radiation is random with respect to the pattern electrodes. Average relative absorption, A average, (effective) adopts an intermediate value between first two curves. When 0.5λ <p <λ, Ar> 90%, the overall effective optimum is located around 8 micrometers steps to maximize the average absorption of a radiation at 10 µm.

It is therefore sufficient that the pattern of the elements conductors be repeated following an understood step typically between λ and λ / 2 to obtain an absorption average greater than 90%. However in the case of infrared detection in the atmospheric band 8-12 micrometers, the rapid decrease in absorption by diffraction effect when p> λ can justify a step of elements: p≤8µm, to avoid the cut-off effect of short wavelengths (8 to 10 µm).

The drawing of the electrodes represented on the Figure 3B, which is a simple network of bands, in step between five micrometers and eight micrometers is complies with this criterion for detection between 8 micrometers and 12 micrometers in a sensitive surface of 50µm x 50µm. We can retain the balanced solution d / e = 1, that is to say fingers of 2.5 micrometers; 2.5 micrometers apart, up to 4 fingers micrometers spaced 4 micrometers apart with a result comparable in terms of absorption around 10 wavelength micrometers.

The equivalent resistance for electrodes drawn in steps of 5 micrometers with p / d = 1 is approximately Rcarré _bolo / 200 whatever the form of the interdigitations, Rcarré _bolo being the layer resistance of the bolometric material. In particular, a comparable value is obtained, apart from peak effects, for the drawing in the form of a "telegraph pole" in FIG. 4A. Such a design does not exhibit any more polarization effect.

In order to optimize the use of the concept proposed, in the detector of the invention, of close coplanar electrodes, we can typically lower the inter-electrode space towards 1 micrometer without being bothered by the non-linearity of the current / voltage characteristic, up to a few volts polarization (Vpol). However Figure 6 shows that it is important to avoid the reasons which present a small step in front of the wavelength according to a direction only, as this results in a polarization.

FIG. 4B presents an exemplary embodiment of "electrodes in a telegraph pole" conforming to this criterion, allowing steps of the order of 2 micrometers or less, without detrimental polarization effect. In a horizontal direction on the diagram, "collectors" are produced with a pitch p1 close to or slightly less than λ so as to ensure the absorption of infrared photons with "horizontal" polarization, relative to the drawing. In the other direction we can have a noticeably smaller step p2, ensuring the absorption of infrared photons with "vertical" polarization. The resulting resistance Ro of a bolometer of 50 µm × 50 µm for a pitch p1 of 2 micrometers with p1 / d # 1 between the electrodes is of the order of Rcarré _bolo / 400 to 500. Even higher TCR materials can then be used at a given resistance R0.

In principle, the same d / e ratio should be observed in both directions if the electrode is obtained in a single operation (a single conductive layer). However, the weakening of the absorption when Rcarré _eq. deviates from the optimal value of around 380 Ω / square is slow and in practice we can tolerate +/- 40% variation without losing more than 10% on absorption. It is therefore easy to adjust the distribution of conductive areas and spaces to best meet the above criteria.

Figures 5A and 5B show examples of second embodiment of the elements conductors of the sensitive part of the detector. These elements are defined according to a plurality of patterns at least two of which are connected to the circuit of reading and constitute strictly speaking the electrodes 13 and 13 ', the other 13 "patterns not being not connected. In Figure 5A the 13 '' patterns are rectilinear conductors, in this example in number of eight, parallel to each other and parallel to electrodes 13 and 13 '. These patterns are separated by one step p such that: λ / 2 <p <λ. In the example of Figure 5B the electrodes 13 and 13 ′ have the shape of a “U”, in which are arranged the 13 "patterns which are also straight conductors, parallel to two outer branches of the "U", the spacing p between these conductors also being: λ / 2 <p <λ. Thus the electrodes 13 and 13 'ensure as in the first embodiment the dual function of electrodes and absorber while the elements 13 '' have only the absorber function. So all characteristics concerning the absorption of conductive elements described above applies also to this embodiment.

This second embodiment allows a greater flexibility in adjusting the Ro resistance due to the possibility of connecting to the circuits of reading two very distant elements as in the Figure 5A or very close together as in the case of Figure 5B. This allows optimization of the coupling with the reading circuit.

Thus, according to this second embodiment represented by FIG. 5A, it is possible to use, under optimal conditions, relatively weak resistive materials, such as for example vanadium dioxide VO ₂ , while retaining the advantages of technological simplicity and immunity to defects of the materials constituting the device specific to the invention. In fact, it is necessary to aim for a high resistance Ro (typically a few MΩ) in order to limit the heating of the detector under the effect of stimuli of reading by the Joule effect, and more simply the overall consumption of the device in operation.

Figure 5B and the various variants that a person skilled in the art will deduce therefrom, makes it possible to carry out resistances of intermediate values according to envisaged optimization objectives. The grounds no connected balance naturally during operation to potentials intermediate between the potentials connected electrodes 13 and 13 '(without disturbing the detector characteristics) according to the principle of divider bridge.

So the fact of following the invention, the infrared absorber functions and electrodes of temperature sensitive material using single conductive layer, significantly simplifies the technological process on the one hand, because only one lithographic level, associated with a single operation of engraving is required. On the other hand a resistance of layer close to 380 Ω / square is obtained with a greater thickness of conductive material, without increase in total mass, which improves the practicality of the deposition process. Indeed if the absorber is made using two continuous layers in parallel, according to a possibility of the prior art, these two layers must have resistance about 750 Ω / square, which requires thicknesses difficult to control with materials even relatively resistive metallic like the titanium nitride: 15 to 40 Angstrom are obtained typically per layer, assuming a resistivity of 100 to 300 µΩcm, which are typical function values of the deposition process. Similarly according to another version of the prior art, which requires a single layer continuous, double thickness is required (30 to 80A in the case of titanium nitride), which is not very practical in most cases. Depending on the solution proposed in the invention, it is necessary to file a thickness close to (ρ / 380). (p / d), i.e. 60 to 160 Angström for d = p / 2, or 90 to 240 Angström for d = p / 3. These thicknesses are reasonably easy to perform and control, proposed p / d values to be considered as typical.

Those skilled in the art also note the advantage of coplanar electrode structures at immunity to short circuits between the two electrodes. Indeed, point integrity faults of the heat-sensitive layer ("pinholes") have no consequence in such a provision. In addition, the presence of an insulated continuous conductive layer electrically of the following heat-sensitive material certain forms of the prior art, makes sensitive also the manufacturing yield at the density of defects in this insulation layer. Now this layer isolation is unnecessary according to the invention. The configuration proposed in the invention is therefore intrinsically robust against the faults of materials used.

Those skilled in the art will also note the particular advantage offered by the forms of the invention as shown in FIG. 5A at level of immunity to short circuits between electrodes. Indeed a random defect at the level of the lithography of the conductive areas will highlight general short circuit two floating elements, without short-circuit the two connected electrodes. It results in a change in resistance Ro, but the elementary point will generally remain operational.

• supporter mécaniquement le corps du dispositif flottant 10 au-dessus du circuit de lecture ;
• connecter électriquement les électrodes réalisées dans la zone I jusqu'aux terminaux métalliques du circuit de lecture, solidaires du substrat 12.

Zones II have two main roles:

• mechanically supporting the body of the floating device 10 above the reading circuit;
• electrically connect the electrodes produced in zone I to the metal terminals of the reading circuit, integral with the substrate 12.

In accordance with the state of the art, these arms of support / insulation can be advantageously cut in the assembly of layers forming the central part I, following a pattern of length L and width l. The thickness is by construction identical to that of zone I. The length of these patterns must be maximum, minimum width and thickness to maximize Rth.

However for reasons of rigidity mechanical, and especially geometric deformations (due to the release of internal constraints from different constituent materials when the structures are freed), we have to find a compromise reasonable. Indeed, it is necessary to control the space between the bolometer body and the reflector to benefit from the quarter-wave effect over the entire detector surface, and the slightest deformation of support arms causes a variation of this space. A lateral cutout of length L equal to one or two sides of an elementary point or pixel represents a maximum at the mechanical level, i.e. from 40 to 80 useful micrometers for a pixel of 50µm × 50µm. The width is typically from 1 to 3 micrometers. Figure 14 shows the easiest drawing of a pixel with two arms of insulation cut along two opposite edges. Each arm in the simplest version consists a conductive layer and a layer of material sensitive.

• supporter mécaniquement l'ensemble I+II ;
• connecter les zones II aux entrées électriques du circuit de lecture ;
• positionner en altitude l'ensemble du dispositif.

Zones III, in accordance with the state of the art, have three main roles:

• mechanically support the assembly I + II;
• connect zones II to the electrical inputs of the read circuit;
• position the entire device at altitude.

According to the preferred embodiment, these zones are made using space pillars reduced, which simply requires opening in zone III. Contact between the electrodes conductive 13 and 13 ', which extends to the area III is mainly provided by a metallic deposit, preferably performed by LPCVD, typically from tungsten, but other materials deposited by LPCVD can be used, such as titanium.

Figures 11 and 12 show cases of connection pillars according to the embodiment preferred using an LPCVD repository. Other modes of possible, for example a deposit metallic by spraying, but in this case it is necessary to provide large diameter connections (several micrometers), with steep sides. This results in a greater loss of space (decrease in the filling factor). The body of the pillar typically consists of several layers for performing the various retail functions of the connection, as explained below.

The most interesting configuration consists of making the pillar by depositing LPCVD with a single sufficiently conductive material (refractory metal typically such as titanium) to simultaneously ensure all technological constraints of contact between the electrode 13 and the metal pad secured to the substrate. Given the resistance very high of the bolometer itself (typically 1MΩ at 100MΩ), it is not necessary to provide contacts very low self resistance (Rc). So titanium or titanium nitride for example are directly adapted to such an embodiment because they have the qualities required compared to other constraints technological assembly. Detailed description which follows supposes the (traditional) implementation of two metallic layers: titanium, tungsten.

The height of the pillar defines the spacing between the sensitive part and the underlying circuit. This spacing is in accordance with the state of the art close to λ / 4 to center the maximum absorption in the vicinity of λ if the surface of the circuit is provided with a infrared reflector.

The manufacturing process of the microbolometer type according to the invention is carried out typically from a read circuit already completed, obtained itself according to known techniques for example microelectronics on substrate of silicon S. The elements appearing on the surface of a such circuit are generally both the studs metallic mounting devices (not shown) for the integration of the final product in a box empty, and the connection pads 20 of each point elementary from the detection device to the points reading circuit correspondents. These studs 20, like mounting pads, are usually reserved at the last metallic interconnection level of the reading circuit. Such a metallic layer is usually passivated using an insulating layer mineral. This mineral layer is engraved on a part of the surface of the pads 20, as well as the pads mounting. The manufacturing process of the the invention can therefore be understood from the structure thus carried out.

The non-detailed steps of the sequence following use well proven techniques known to those skilled in the art.

The process steps of an example of manufacturing are organized according to the following sequence:

1/ Première étape (figure 8) : dépôt métallique pour réaliser le réflecteur (21). 200A à 500A de titane et 1000A d'aluminium, par exemple, déposés par pulvérisation suffisent.

2/ Seconde étape : définition à l'aide d'un premier niveau lithographique des motifs du réflecteur et gravure de ce dernier selon les techniques connues de l'homme de l'art.

3/ Troisième étape : étendage d'une couche de polyimide (22), et recuit de cette couche, à une température de l'ordre de 400 à 450°C L'épaisseur après recuit vaut environ 2.5 micromètres en moyenne pour assurer l'effet cavité résonante, conformément à l'état de l'art (la topographie du circuit de lecture induit une dispersion d'épaisseur, peu critique). Ce recuit est nécessaire pour éviter la modification des propriétés du polyimide pendant la suite du procédé. La température maximale atteinte pendant ce recuit doit être au moins égale à la température maximale atteinte par la suite jusqu'à la réalisation du détecteur.

4/ Quatrième étape (figure 9) : dépôt, typiquement par pulvérisation réactive, d'une couche de nitrure de titane 23 et d'une couche d'aluminium 24. Le nitrure de titane constitue les électrodes du bolomètre (car la couche d'aluminium est éliminée ultérieurement). Son épaisseur doit donc être adaptée à la fonction absorbeur optique, conformément à l'objet de l'invention. Le choix du nitrure de titane est avantageux pour plusieurs raisons : il forme des contacts de bonne qualité avec le silicium amorphe ; il est relativement facile à déposer en couches très minces ; il présente une résistivité plutôt élevée, et à la rigueur ajustable ; et, surtout, il est facile de graver très sélectivement une couche d'aluminium déposée au-dessus. Pour une résistivité typique de 130 à 140µΩcm, on dépose environ 75 Angström de nitrure de titane.

5/ Cinquième étape : masquage à l'aide d'un second niveau lithographique des ouvertures de contact vers le circuit de lecture, qu'on appelle "vias", et gravure de la couche d'aluminium, de la couche de nitrure de titane, et de la couche de polyimide jusqu'à la base de celle-ci selon les techniques connues de l'homme de l'art.

6/ Sixième étape (figure 10) : dépôt, typiquement par pulvérisation cathodique, de titane (25). Le rôle du titane est d'assurer un contact électrique de bonne qualité entre l'aluminium du réflecteur d'une part et l'aluminium en surface du polyimide d'autre part et le matériau de remplissage des vias (tungstène).

7/ Septième étape : dépôt par LPCVD d'une couche 26 de 5000 Angström à 1 micromètre de tungstène selon les techniques connues de l'homme de l'art. Une variante avantageuse de réalisation des métallisations consiste à déposer directement et uniquement du titane par LPCVD, en lieu et place des sixième et septième étapes. En effet la résistance des bolomètres est très grande devant la résistance propre du contact et du pilier de connexion ainsi réalisé par les couches 25, 26, et il n'y a pas lieu de recourir à des techniques complexes à deux couches successives conformes à l'art traditionnel en la matière.

8/ Huitième étape (figure 11) : définition des connexions métalliques à l'aide d'un troisième niveau lithographique. Typiquement la surface protégée par la résine photosensible recouvre les ouvertures précédentes de 1 à 2 micromètres, afin d'assurer la continuité électrique métallique entre l'électrode de nitrure de titane et l'entrée du circuit de lecture, via le pilier ainsi constitué.

9/ Neuvième étape : gravure du tungstène, et du titane, ou de la couche unique de titane suivant la variante énoncée plus haut puis gravure de l'aluminium de l'étape 4, par les procédés usuels de gravure. On effectue le retrait de la résine photosensible par les moyens usuels.

10/ Dixième étape (figure 12) : définition, à l'aide d'un quatrième niveau lithographique, de la couche de nitrure de titane. Cette couche constitue les électrodes du bolomètre. Le dessin du masque doit donc être conforme aux principes de dessin énoncés précédemment, qui n'impliquent pas de définition lithographique fine. Des moyens peu sophistiqués suffisent (lithographie par proximité/contact) pour l'optimisation dans la bande 8-12µm. Après cette gravure la résine photosensible est éliminée par les moyens usuels. Suivant la configuration préférée, on définit des électrodes interdigitées de 2.5 à 4 micromètres de largeur, espacées de 2.5 à 4 micromètres, suivant un réseau parallèle simple.

11/ Onzième étape (figure 13) : dépôt du matériau bolométrique (27), typiquement, mais pas exclusivement, du silicium amorphe dopé. Ce type de dépôt relève de l'art connu en ce qui concerne l'optimisation et le contrôle de la résistivité en fonction du mélange de gaz et des autres paramètres de dépôt. La résistivité du matériau doit être adaptée au dessin des électrodes, ou réciproquement le dessin des électrodes peut être adapté à la résistivité maximale que l'on sait reproduire. Dans la configuration préférée, on dépose typiquement 1000 Angström de silicium amorphe PECVD de type p à 200 KΩcm de résistivité, en mettant en oeuvre le mélange de gaz adéquat.

12/ Douzième étape (figure 14) : définition et gravure du ( ou des) point(s) élémentaire(s) et des bras d'isolation thermique à l'aide d'un cinquième et dernier niveau lithographique selon les techniques de l'homme de l'art.

13/ Treizième étape : libération des microstructures, typiquement à l'aide d'une gravure sèche oxydante telle que pratiquée par l'homme de l'art, pour l'élimination sèche des résines photosensibles.

(It should be noted that on the top views described below, the solid lines correspond to the contours of the lithographic level considered while the dotted lines represent one or more previous levels).

1 / First step (Figure 8): metallic deposit to make the reflector (21). 200A to 500A of titanium and 1000A of aluminum, for example, deposited by spraying are sufficient.

2 / Second step: definition using a first lithographic level of the patterns of the reflector and etching of the latter according to techniques known to those skilled in the art.

3 / Third stage: spreading of a layer of polyimide (22), and annealing of this layer, at a temperature of the order of 400 to 450 ° C. The thickness after annealing is worth approximately 2.5 micrometers on average to ensure the resonant cavity effect, in accordance with the state of the art (the topography of the reading circuit induces a dispersion of thickness, not very critical). This annealing is necessary to avoid modification of the properties of the polyimide during the rest of the process. The maximum temperature reached during this annealing must be at least equal to the maximum temperature reached thereafter until the detector is produced.

4 / Fourth step (FIG. 9): deposition, typically by reactive sputtering, of a layer of titanium nitride 23 and of a layer of aluminum 24. Titanium nitride constitutes the electrodes of the bolometer (because the layer of aluminum is removed later). Its thickness must therefore be adapted to the optical absorber function, in accordance with the object of the invention. The choice of titanium nitride is advantageous for several reasons: it forms good quality contacts with amorphous silicon; it is relatively easy to deposit in very thin layers; it presents a rather high resistivity, and with an adjustable rigor; and, above all, it is easy to very selectively etch an aluminum layer deposited on top. For a typical resistivity of 130 to 140µΩcm, approximately 75 Angstroms of titanium nitride are deposited.

5 / Fifth step: masking with a second lithographic level of the contact openings towards the reading circuit, which is called "vias", and etching of the aluminum layer, of the titanium nitride layer , and from the polyimide layer to the base thereof according to techniques known to those skilled in the art.

6 / Sixth step (FIG. 10): deposition, typically by sputtering, of titanium (25). The role of titanium is to ensure good quality electrical contact between the aluminum of the reflector on the one hand and the aluminum on the surface of the polyimide on the other hand and the filling material of the vias (tungsten).

7 / Seventh step: deposit by LPCVD of a layer 26 of 5000 Angstroms at 1 micrometer of tungsten according to techniques known to those skilled in the art. An advantageous variant of metallization consists in depositing titanium directly and only by LPCVD, in place of the sixth and seventh steps. In fact, the resistance of the bolometers is very large compared to the inherent resistance of the contact and of the connection pillar thus produced by the layers 25, 26, and there is no need to resort to complex techniques with two successive layers in accordance with the traditional art on the subject.

8 / Eighth step (figure 11): definition of the metallic connections using a third lithographic level. Typically the surface protected by the photosensitive resin covers the preceding openings of 1 to 2 micrometers, in order to ensure metallic electrical continuity between the titanium nitride electrode and the input of the reading circuit, via the pillar thus formed.

9 / Ninth step: etching of tungsten, and of titanium, or of the single layer of titanium according to the variant set out above, then etching of the aluminum of step 4, by the usual etching processes. The photosensitive resin is removed by the usual means.

10 / Tenth step (Figure 12): definition, using a fourth lithographic level, of the titanium nitride layer. This layer constitutes the electrodes of the bolometer. The drawing of the mask must therefore comply with the drawing principles stated above, which do not imply a fine lithographic definition. Unsophisticated means are sufficient (proximity / contact lithography) for optimization in the 8-12 µm band. After this etching, the photosensitive resin is removed by the usual means. According to the preferred configuration, interdigitated electrodes 2.5 to 4 micrometers wide are defined, spaced 2.5 to 4 micrometers apart, according to a simple parallel network.

11 / Eleventh step (Figure 13): deposition of the bolometric material (27), typically, but not exclusively, doped amorphous silicon. This type of deposition falls within the known art with regard to the optimization and the control of the resistivity as a function of the gas mixture and of the other deposition parameters. The resistivity of the material must be adapted to the design of the electrodes, or conversely the design of the electrodes can be adapted to the maximum resistivity which it is known to reproduce. In the preferred configuration, 1000 Angstroms of P-type PECVD amorphous silicon are typically deposited at 200 KΩcm of resistivity, using the appropriate gas mixture.

12 / Twelfth step (figure 14): definition and etching of the elementary point (s) and thermal insulation arms using a fifth and last lithographic level according to the techniques of skilled in the art.

13 / Thirteenth step: release of the microstructures, typically using an oxidizing dry etching as practiced by those skilled in the art, for the dry removal of photosensitive resins.

In practice it is necessary to perform the functionality test or sort and especially the cutting of the components before releasing the structures, for reasons of fragility and contamination during these different stages. After release, devices are vacuum mounted in an enclosure, temperature-adjustable, comprising the elements optics required.

We will now consider successively several variants of the device of the invention.

Titanium nitride and amorphous silicon usually have a level of constraints high internal. The simplest layout offered in two superimposed layers (configuration called "preferred") can lead to deformations important, after the release stage, by effect "bimetallic strip".

It is in this case advantageous to compensate this effect by judiciously arranging layers adequate on both sides of the conductive elements. Thus, as explained above, we can file the sensitive material on both sides of the elements conductors; we can also deposit on one side conductive elements the sensitive material and on the other side a layer of passive material. Finally, we can also deposit on the face (s) of the material sensitive not in contact with conductive elements a layer of passive material.

The so-called passive material is preferably a material with low thermal conductivity because these diapers are also found in the arms of isolation. So we preferentially use silicon oxide (SiO), silicon nitride (SiN) or of amorphous silicon deposited according to the traditional methods.

We can thus without significantly complicating the process deposit for example a layer 30 from 100 to 200 Angstrom of silicon oxide or nitride of silicon or amorphous silicon on the surface of the polyimide, before deposition of the nitride electrode titanium, or / and an identical layer 31 after depositing the sensitive material. These two layers are engraved with the others during the twelfth stage.

The preferred configuration is a stack symmetrical which effectively stabilizes the structure, as shown in Figure 15. An advantage associated with this arrangement is the effect of electrical passivation non-metallized surfaces of sensitive material or not opposite the electrodes.

The additional mass can simply be offset by an equivalent decrease in the layer of sensitive material, for example silicon amorphous. Indeed, the densities and capacities intrinsic calories from silicon nitride, silicon oxide and amorphous silicon are not very different, no more than their conductivities respective thermal.

Another advantageous configuration is achievable, always with the aim of improving the mechanical stability of the released structures, and possibly the electrical stability of the bolometers. It consists of a symmetrical distribution of the layers active (and possibly passive) on both sides of the titanium nitride electrode, in accordance with the figure 16.

Just file in the most case sophisticated where we pass the silicon surfaces amorphous on the surface of the polyimide (between the first step and the second step) for example a layer 32 of 100 to 200 Angstroms of silicon nitride, then a 300 to 400 Angstrom layer of amorphous silicon weakly doped. During the eleventh step we repeat this sequence, but reversed, i.e. 300 to 400 Lightly doped silicon angstrom (34), then 100 to 200 Angstroms of silicon nitride (35).

All the stacked layers are then engraved during the twelfth stage, as before. The total mass is similar to the stack of reference, but the structure is particularly stable by construction, because perfectly symmetrical. Additionally the central titanium nitride electrode makes contact of amorphous silicon on its two faces, which goes in the direction of improving injecting current through the structure.

Figures 8 to 16, which illustrate the process of the invention represent sectional views of the structure produced, as well as views of lithographic levels, at the different stages of said process or variants thereof.

FIG. 8A illustrates a section AA ' represented in FIG. 8B after definition of the reflector 21.

FIG. 9A illustrates a section AA ' represented in FIG. 9B after definition of the connection openings 30.

Figure 10 illustrates a section AA 'after interconnection metallization deposit.

FIG. 11A illustrates a section AA ' illustrated in FIG. 11B after definition of metallic interconnection; reservations metal 31 being shown in Figure 11B.

FIG. 12A illustrates a section AA ′ illustrated in FIG. 12B after definition of the electrodes with a tungsten thickness more important ("plug" configuration); the reference 32 on FIG. 12B illustrating an open pattern (engraved area).

Figure 13 illustrates a section AA 'after deposit of bolometric material.

FIG. 14A illustrates a section AA ' shown in Figure 14B after crosslinking.

FIG. 15 illustrates a section AA ′ of a passivated structure (first variant).

Figure 16 illustrates a structure passivated symmetrical (second variant).

Claims (24)

Infrared detector, comprising: a sensitive part comprising a sensitive element comprising a layer of material whose resistivity varies with temperature, conductive elements (13, 13 ′) for ensuring the functions of electrodes of said detector and of infrared absorber; at least one support element (11) of the sensitive part able to position said sensitive part, to connect it electrically to a reading circuit and to thermally isolate it from the rest of the structure; characterized in that all the conductive elements are arranged on the same face of the layer of temperature-sensitive material. Detector according to claim 1, characterized in that the conductive elements have two interdigitated electrodes realizing simultaneously the functions of electrodes and absorber, these electrodes being connected to the reading. Detector according to claim 1, characterized in that the conductive elements have two electrodes of a first type connected to the read circuit and at least one electrode of a second type brought to a floating potential, the electrode of the second type ensuring only the function absorber and electrodes of the first type ensuring simultaneously the functions of electrodes and absorber. Detector according to any one of previous claims, characterized in that at at least two conductive elements ensure the connection electric to the reading circuit. Detector according to any one of previous claims, characterized in that said face of the layer of sensitive material is the underside, the conductive elements being arranged under said layer. Detector according to any one of claims 1 to 4, characterized in that said face of the sensitive material layer is the face upper, the conductive elements being arranged on said layer. Detector according to any one of claims 1 to 4, characterized in that the conductive elements are inserted between two layers of sensitive material. Detector according to claim 1, characterized in that the material sensitive to temperature is amorphous silicon or oxide vanadium. Detector according to claim 1, characterized in that the free space defined between the conductive elements (13, 13 ') consists only of temperature sensitive material. Detector according to claim 1, characterized in that a reflector (21) is provided in surface of the substrate supporting the reading circuit, and located at λ / 4 below the plane of the sensitive element, λ being the typical wavelength of the radiation received. Detector according to claim 1, characterized in that the pattern of the conductive elements is repeated in a step typically between λ and λ / 2 to obtain a higher average absorption 90%. Detector according to claim 1, characterized in that each support element comprises at least one connecting pillar and at least one arm. Infrared detector, characterized in that it consists of a set of detectors according to any one of claims 1 to 12. Method for manufacturing an infrared detector according to any one of the preceding claims, characterized in that, starting from a reading circuit already completed from a substrate, it comprises the following steps: a) formation of a sacrificial layer; b) depositing at least one conductive layer for producing the conductive elements; c) making at least one contact opening towards the reading circuit by localized etching of the conductive layer and the sacrificial layer; d) depositing at least one conductive material and etching the latter so as to ensure the electrical connection between the conductive elements and the reading circuit, the pattern thus produced forming the pillar of the support; e) etching the conductive layer deposited in step b) so as to define the patterns of the conductive elements; f) depositing on the assembly, the layer of temperature-sensitive material and etching of the various layers arranged above the sacrificial layer so as to define the sensitive element and the arm of the support; g) etching of the sacrificial layer so as to release the sensitive part and the pillar. Method according to claim 14, characterized in that step f) of material deposition sensitive is carried out before step b), the material sensitive then being etched in step c). Method according to claim 14, characterized in that the sensitive material is deposited in a first layer before step b) and in one second layer in step f), the first layer of sensitive material then being etched in step c). Method according to claim 14, characterized in that step d) is carried out after step b). Method according to claim 14, characterized in that it is carried out prior to the step a), depositing and etching a metal layer for form a reflector. Method according to claim 14, characterized in that the sacrificial layer is a polyimide layer which has been subjected to annealed. Method according to claim 14, characterized in that the conductive layer of step b) is a bilayer of titanium nitride and aluminum, the conductive material of step d) being then deposited after completion of steps b) and c), this conductive material and aluminum then being engraved in the same way in step d). Method according to claim 14, characterized in that the conductive material of step d) is a multilayer of which at least one of the layers is formed by LPCVD. Method according to claim 14, characterized in that the conductive elements having first and second faces, the sensitive material being in contact with one of the faces of said elements, the other face is brought into contact with a layer of passive material. Method according to claim 14, characterized in that a layer of passive material is deposited on the face (s) of the sensitive material, not in contact with the conductive elements. The method of claim 14 or 23, characterized in that the passive material is chosen among silicon oxide, silicon nitride and amorphous silicon.

Images